Max power dissipation for 2N3055 without heatsink?

[Bill Bowden]

According to a reference the 2N3055 is good for 5.8 watts without a heatsink at 25c ambient temp, and the case temp rise is 30 degrees per watt. So, if ambient is 44C and power is 2 watts, case temp should be 104C, or boiling point of water. Any problem doing this, or should a heat sink be used?

 

I kinda wonder why. the whole idea of the TO-3 case style is to facilitate heat sinking. As far as space and cost go, a smaller transistor with a heat sink would be better in alot of ways.

Of course people have their reasons, maybe you just have some laying around ?

As the specs say, of course you CAN? get the transistor that hot. Just make sure you compensate properly if using it in the linear range. Of course as a switcher you will have to design the drive to accomodate the degraded switching characteristics at the temperature. With switching however, it probably doesn't need to track, just optimize it for 104C and it should be fine at 25C. Just watch that reverse B-E voltage.

With the surface area of the TO-3 case style, even putting it in contact with a circuit board will help, so it might not be totally zero heat sink, other than the surface area of the transistor itself. Plus, someting that heavy, you might want to screw it down somehow, so that should help almost no matter what you screw it to. Well maybe certain types of rubber, or asbestos or something. I odubt you would be looking specifically for a thermal insulator unless you have one wierd application there...

 

[Cydrome Leader]

Bill Bowden <bperryb@bowdenshobbycircuits.info> wrote:
> According to a reference the 2N3055 is good for 5.8 watts without a heatsink at 25c ambient temp, and the case temp rise is 30 degrees per watt. So, if ambient is 44C and power is 2 watts, case temp should be 104C, or boiling point of water. Any problem doing this, or should a heat sink be used?

keep in mind datasheets should not always be read literally.

Take a to-3 cased transistor. It's not even possible to use it with
nothing attached to the case as that's an electrical connection. So while,
yes you can probably dissipate 5.8w from one, it's not a real scenario
that will even happen where the transistor is just floating in air with
only the two leads connected to anything.

So as in the other followup post, even connecting the thing to a PCB will
add some heatsinking and is probably fine.

 

[Cydrome Leader]

RobertMacy <robert.a.macy@gmail.com> wrote:

On Thu, 24 Apr 2014 18:30:40 -0700, Bill Bowden
bperryb@bowdenshobbycircuits.info> wrote:

According to a reference the 2N3055 is good for 5.8 watts without a
heatsink at 25c ambient temp, and the case temp rise is 30 degrees per
watt. So, if ambient is 44C and power is 2 watts, case temp should be
104C, or boiling point of water. Any problem doing this, or should a
heat sink be used?


Historically, the 2N3055 was the workhorse transistor for the linear power
supplies of yesteryear. Almost every 10, 20, 100W multiple output, or even
up to the 300W supplies used these as the highly heatsinked linear
regulated output. Even HP's lab bench supplies used them as the final
regulators.

Even with the linear power supplies of today, they still use the 2n3055.
Even the knock off ones do. In fact, I can't think of the last time I saw
a to-3 cased transistor that wasn't a 2n3055.

 

[RobertMacy]

On Thu, 24 Apr 2014 18:30:40 -0700, Bill Bowden
<bperryb@bowdenshobbycircuits.info> wrote:

According to a reference the 2N3055 is good for 5.8 watts without a
heatsink at 25c ambient temp, and the case temp rise is 30 degrees per
watt. So, if ambient is 44C and power is 2 watts, case temp should be
104C, or boiling point of water. Any problem doing this, or should a
heat sink be used?

Historically, the 2N3055 was the workhorse transistor for the linear power
supplies of yesteryear. Almost every 10, 20, 100W multiple output, or even
up to the 300W supplies used these as the highly heatsinked linear
regulated output. Even HP's lab bench supplies used them as the final
regulators.

I have NEVER found an equivalent transistor component. The Tjmax is 150C
where most transistors have Tjmax in the 125C range. That extra 25C is
where the allure of 2N3055 lies. The MOST important spec to never violate
is the junction temperature. As you probably know, if you violate that
maximum, you are 'baking' the NPN back into the ovens, the dopants
continue to diffuse, and the beta goes to hell in a handbasket as the base
junction just goes??? You can lose beta, or even end up connecting the
collector region to the emitter region, which means, no more transistor.

So ALWAYS calculate Tj, keep below max, and assume NO margin and you will
be alright. I know, I know, you have no control over the thermal
conductivity between junction and case, so why not only 'think' case temp?
Well, thinking junction temp keeps your mind concentrating on the
important issue, not a 'sideways' observation. And, you'll know the true
origin of the limit. And, if you have a spate of failures, can help you
look for the failure mode, as an example, like when the IC house mounted
the chip, they didn't do it right this run, and there were little voids
all over underneath which destroyed the chip to case thermal conductivity,
yet didn't touch the chip to case electrical conductivity. IMO, the way to
think is ALWAYS to think in terms of the rudimentary, not in terms of the
'outsides' or 'secondary' terms.

PS: don't assume ambient air is the same everywhere. in a box ambient is
not quite so ambient. at least without a fan.

 

[Bill Bowden]

On Friday, April 25, 2014 7:37:55 AM UTC-7, RobertMacy wrote:

On Thu, 24 Apr 2014 18:30:40 -0700, Bill Bowden


According to a reference the 2N3055 is good for 5.8 watts without a
heatsink at 25c ambient temp, and the case temp rise is 30 degrees per
watt. So, if ambient is 44C and power is 2 watts, case temp should be
104C, or boiling point of water. Any problem doing this, or should a
heat sink be used?


Historically, the 2N3055 was the workhorse transistor for the linear power
supplies of yesteryear. Almost every 10, 20, 100W multiple output, or even
up to the 300W supplies used these as the highly heatsinked linear
regulated output. Even HP's lab bench supplies used them as the final
regulators.

I have NEVER found an equivalent transistor component. The Tjmax is 150C
where most transistors have Tjmax in the 125C range. That extra 25C is
where the allure of 2N3055 lies. The MOST important spec to never violate
is the junction temperature. As you probably know, if you violate that
maximum, you are 'baking' the NPN back into the ovens, the dopants
continue to diffuse, and the beta goes to hell in a handbasket as the base
junction just goes??? You can lose beta, or even end up connecting the
collector region to the emitter region, which means, no more transistor.

So ALWAYS calculate Tj, keep below max, and assume NO margin and you will
be alright. I know, I know, you have no control over the thermal
conductivity between junction and case, so why not only 'think' case temp?
Well, thinking junction temp keeps your mind concentrating on the
important issue, not a 'sideways' observation. And, you'll know the true
origin of the limit. And, if you have a spate of failures, can help you
look for the failure mode, as an example, like when the IC house mounted
the chip, they didn't do it right this run, and there were little voids
all over underneath which destroyed the chip to case thermal conductivity,
yet didn't touch the chip to case electrical conductivity. IMO, the way to
think is ALWAYS to think in terms of the rudimentary, not in terms of the
'outsides' or 'secondary' terms.

PS: don't assume ambient air is the same everywhere. in a box ambient is
not quite so ambient. at least without a fan.

This link claims the Tjmax is 200c and junction to case resistance is 1.52 per watt. So, at 2 watts the junction should only be about 3 degrees higher than the case. Seems low to me. I'm going to try a double sided 5 square inch copper board which should increase the area about 3X. All the other components will fit on the board. It's a shunt regulator for a small solar panel and there isn't any headroom for a series regulator. I wonder how much extra thermal resistance will be added if I spray the finished board with Krylon?

http://www.morpheustechnology.com/ebooks/eXe_Power%20Supplies/Switch_Mode_Power_Supplies/thermal.html

 

Actually there should be plenty of headroom for a MOSFET.

Shunt regulators are so inefficient. Actually you could have a switched mode shunt regulator but if there is to be any filtering after it, there will have to be a diode to block the post filter's current from the shunt regulator when it is switched on.

Another plane would be to use a ballast. you should know the maximum current the shunt needs to pul to regulate. Use a resistor that will shunt it enough and shift the dissipatin to the resistor. At that point you can switch mode it if you feel so inclined, but you still aren't going to get very high efficiency.

 

[Phil Allison]

"Bill Bowden"


This link claims the Tjmax is 200c

** That is the damage temp, never to be exceeded.

The max safe working case temp is more like 100C.


and junction to case resistance is 1.52 per watt.
So, at 2 watts the junction should only be about 3 degrees higher than the
case. Seems low to me.


** Many TO3s are better, down to 0.6C per watt for 200W rated BJTs.

Use one of these, takes hardly any PCB space.

http://www.westfloridacomponents.com/G522APF06/TO3+Heatsink+CTS+UP1-TO3-CB.html


..... Phil

 

[RobertMacy]

On Fri, 25 Apr 2014 18:01:37 -0700, Bill Bowden
<bperryb@bowdenshobbycircuits.info> wrote:

...snip...

This link claims the Tjmax is 200c and junction to case resistance is
1.52 per watt. So, at 2 watts the junction should only be about 3
degrees higher than the case. Seems low to me. I'm going to try a double
sided 5 square inch copper board which should increase the area about
3X. All the other components will fit on the board. It's a shunt
regulator for a small solar panel and there isn't any headroom for a
series regulator. I wonder how much extra thermal resistance will be
added if I spray the finished board with Krylon?

http://www.morpheustechnology.com/ebooks/eXe_Power%20Supplies/Switch_Mode_Power_Supplies/thermal.html

Right! It's 200C, and EVERYTHING else is 150C!! Some day will check facts
and quit relying on memory. NOT!

To understand air cooling, someone once told me to think in terms of a
'stiction' layer of air. Air movement takes all the heat you can give it,
but first the heat must travel through that 'stiction' layer. Easy to
calculate the C/W based upon area and conductivity through that layer. A
fan blowing across the component only makes the stiction layer thinner,
that's why the curves don't go to zero [C/W], because there's ALWAYS a
layer of air there. Don't know how accurate that mental model is, but
certainly helps envision why it's so difficult to cool a component in air.

To find the effect of the Krylon coating, bet you can do a 'layer'
calcuation. just like stacking dielectrics in a capacitor. Assume no
'hotspots', then have thermal conductivity thru Krylon to surface, then
thermal conductivity into the air. Two terms in a row. Not much effect,
but will be there.

 

[Jim Thompson]

On Sat, 26 Apr 2014 06:14:23 -0700, RobertMacy
<robert.a.macy@gmail.com> wrote:

On Fri, 25 Apr 2014 18:01:37 -0700, Bill Bowden
bperryb@bowdenshobbycircuits.info> wrote:

...snip...

This link claims the Tjmax is 200c and junction to case resistance is
1.52 per watt. So, at 2 watts the junction should only be about 3
degrees higher than the case. Seems low to me. I'm going to try a double
sided 5 square inch copper board which should increase the area about
3X. All the other components will fit on the board. It's a shunt
regulator for a small solar panel and there isn't any headroom for a
series regulator. I wonder how much extra thermal resistance will be
added if I spray the finished board with Krylon?

http://www.morpheustechnology.com/ebooks/eXe_Power%20Supplies/Switch_Mode_Power_Supplies/thermal.html


Right! It's 200C, and EVERYTHING else is 150C!! Some day will check facts
and quit relying on memory. NOT!

To understand air cooling, someone once told me to think in terms of a
'stiction' layer of air. Air movement takes all the heat you can give it,
but first the heat must travel through that 'stiction' layer. Easy to
calculate the C/W based upon area and conductivity through that layer. A
fan blowing across the component only makes the stiction layer thinner,
that's why the curves don't go to zero [C/W], because there's ALWAYS a
layer of air there. Don't know how accurate that mental model is, but
certainly helps envision why it's so difficult to cool a component in air.

To find the effect of the Krylon coating, bet you can do a 'layer'
calcuation. just like stacking dielectrics in a capacitor. Assume no
'hotspots', then have thermal conductivity thru Krylon to surface, then
thermal conductivity into the air. Two terms in a row. Not much effect,
but will be there.

Depending on application, thermal _capacity_ can matter more than
thermal _resistance_.

For the TOW missile voltage regulator I used a steel TO-3 package with
no heat sink. Its thermal capacity held temperature rise down long
enough to reach the target without excessive temperature rise... after
that it doesn't matter >:-}

...Jim Thompson
--
| James E.Thompson | mens |
| Analog Innovations | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| San Tan Valley, AZ 85142 Skype: Contacts Only | |
| Voice480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |

I love to cook with wine. Sometimes I even put it in the food.